The homogeneity and Ar-dating suitability of the GL-O reference material were re-evaluated to determine whether this material is sufficiently homogeneous to be suitable for the calibration of modern high sensitivity instruments. Based on new micro-analyses and noble gas determinations, our contribution reveals several kinds of inhomogeneity at the grain scale: disparity in the glauconitisation among and within the pellets, variable occurrence of a phosphatic component within pellets (1% m/m on average), and rare occurrences of calcite and detrital grains. Measurements on test portions of ≤ 1 mg reflect such heterogeneity with variability in 40 Ar* content that exceeds analytical uncertainty, including a few highly anomalous values. The lesser evolved glauconite population yielded 40 Ar* contents ~15% lower than the value of 24.8 nl g -1 recommended by Odin et al. (1982, Numerical dating in stratigraphy. Wiley (Chichester, UK), 123-148). But the measured concentrations of 40 Ar* converge towards the aforementioned value as test portion mass increased to > 3 mg. A few rare 3 mg experiments still yielded 40 Ar* contents lower than the recommended value (down to 24.0 nl g -1 ), and we recommend using more conservative minimum masses of 5-10 mg. A further purification step for GL-O or the intercalibration of its powder version could be considered to diminish the size of the test portions and the intensity of the measured signals.